Scanning Microscopy Scanning Microscopy

Volume 1 Number 2 Article 33

3-20-1987

Utilization of Immunogold Labeling to Compare the Adsorption Utilization of Immunogold Labeling to Compare the Adsorption

Behavior of Fibrinogen, Fibronectin and Albumin on Polymers Behavior of Fibrinogen, Fibronectin and Albumin on Polymers

K. D. Murthy University of Wisconsin

A. R. Diwan University of Wisconsin

S. R. Simmons University of Wisconsin

R. M. Albrecht University of Wisconsin

S. L. Cooper University of Wisconsin

Follow this and additional works at: https://digitalcommons.usu.edu/microscopy

Part of the Life Sciences Commons

Recommended Citation Recommended Citation Murthy, K. D.; Diwan, A. R.; Simmons, S. R.; Albrecht, R. M.; and Cooper, S. L. (1987) "Utilization of Immunogold Labeling to Compare the Adsorption Behavior of Fibrinogen, Fibronectin and Albumin on Polymers," Scanning Microscopy: Vol. 1 : No. 2 , Article 33. Available at: https://digitalcommons.usu.edu/microscopy/vol1/iss2/33

This Article is brought to you for free and open access by the Western Dairy Center at DigitalCommons@USU. It has been accepted for inclusion in Scanning Microscopy by an authorized administrator of DigitalCommons@USU. For more information, please contact digitalcommons@usu.edu.

Scanning Microscopy, Vol. 1, No. 2, 1987 (Pages 765-773) 0891-7035/87$3.00t-.00 Scanning Microscopy International, Chicago (AMF O'Hare), IL 60666 USA

UTILIZATION OF IMMUNOGOLD LABELING TO COMPARE THE ADSORPTION BEHAVIOR OF FIBRINOGEN, FIBRONECTIN AND ALBUMIN ON POLYMERS

K. D. Murthy(!), A. R. Diwan(l), S. R. Simmons(2), R. M. Albrecht (2) S. L. Cooper*(l)

Department of Chemical Engineeri·ng(l) and Veterinary Sciencesl2 University of Wisconsin Madison, Wisconsin 53706

(Received for publication November 12, 1986, and in revised form March 20, 1987)

Abstract

Immunogold labeling followed by scanning electron microscopy (SEM) was used to examine the surface distribution of adsorbed plasma proteins. Adsorption was performed under various conditions on six different polymers; [low density polyethylene (PE), chromic acid­oxidized PE (OXPE), solution grade Biomer® (SB), Teflon-(FEP)®, a laboratory synthesized polyurethane containing some zwitterions (ZW) and a polydimethylsiloxane based polyurethane (ZS) also containing zwitterions]. The pro­teins used were purified human and canine fibrinogen, fibronectin, and serum albumin. The immunogold staining technique was success­ful in the labeling of the adsorbed proteins. The adsorbed proteins were distributed dif­ferently on the polymers selected. Human and canine fibrinogen were found to cover all sur­faces in a dense, uniform fashion. Albumin covered most surfaces in a less uniform fashion and on the zwitterionomers covered only a por­tion of the surface, leaving large bare patches. Fibronectin appeared to deposit unevenly, forming a network on part of the sur­face and uniformly coating other parts.

KEY WORDS: Adsorption, albumin, antibody, antigen, conformation, fibrinogen, fibronectin, immunogold, interfacial, polymers

*Address for correspondence: S.L. Cooper Dept. of Chemical Engineering University of Wisconsin Madison, Wisconsin 53706

Phone No. (608) 262-1092

765

Introduction

The study of protein adsorption at solid­liquid interfaces is very important for the understanding of several phenomena such as biofouling, receptor-ligand interactions, and artificial surface-induced thrombosis [1,14]. It has been demonstrated that the adsorption of proteins to polymers upon blood exposure is an event of primary importance in determining the subsequent thromboembolic events [3]. The physicochemical characteristics of the surface (e.g., functional groups, hydrophobicity/ hydrophilicity and surface energy [2,10], surface charge density [26], microcrystallin­ity, microphase separation, etc.) and the physical and chemical nature of the protein molecule (e.g., molecular weight, size and shape, charge and primary, secondary and tertiary structure [1,14], availability of reactive chemical groups such as thiols, etc.) together determine the adsorption behavior of a protein onto a given surface. The adsorption behavior may be characterized by several para­meters. Some of the parameters usually studied include the amount of protein adsorbed, the reversibility of adsorption, and the rate of adsorption and desorption processes, and, to a lesser extent, the modification of biological activity due to adsorption. Another important aspect is the surface distribution of the adsorbed protein at the microscopic level. The distribution may reveal the nature of interac­tion between the surface and the adsorbed protein, as well as the possible biochemical interactions between adsorbed protein molecules, which may be brought about simply by their proximity on surface, or by the specific chemical nature of the surface, which creates conducive microenvironmental conditions. The conformation and biological activity of the proteins may be altered by these interactions.

Despite the importance of the nature of surface distribution of proteins, very little attention has been paid to its study [18,19, 23,24]. Surface distribution can be analyzed by techniques such as immunoferritin transmission electron microscopy (TEM),

K. D. Murthy et al.

modified negative staining electron microscopy and by partial gold decoration TEM [17]. The first two techniques require surface replica­tion procedures, which may introduce several artifacts in the final image, and the last technique lacks sufficient resolution and sensitivity [17]. Further, the TEM techniques are not optimal for studies of protein adsorp­tion on many surfaces of biomedical importance due to the difficulties of preparing sufficiently thin polymer films representative in properties of the bulk polymer surface.

In contrast, immunogold staining is a use­ful technique for the direct visualization of the distribution of adsorbed proteins on surfaces. Non-cross-reactive antibodies, specific to a selected protein, can be chosen. The technique can be used in single as well as competitive and sequential protein adsorption studies. Non-specific adsorption of immunogold particles onto bare polymer surfaces is minimal. The colloidal gold-antibody solutions are easy to prepare and use and the size of the gold particles can be easily controlled. We have chosen 18 nm diameter gold particles because they can be easily distinguished using conventional secondary electron imaging. Backscattered electron imaging (BSE) can be used to unambiguously identify gold particles although the higher accelerating voltages used to facilitate BSE may be detrimental to certain polymers. High resolution scanning electron microscopy would be the method of choice for observation of the gold label, the polymer, and the protein coat; however, this instrumentation is presently not generally available.

The proteins used for this study, human (HFGN) and canine fibrinogen (CFGN), human (HFN) and canine fibronectin (CFN), human (HSA) and canine serum albumin (CSA) possess substan­tially different physical and biochemical properties. Differences exist in their molecular weights, primary, secondary and tertiary structures and molecular dimensions, reactive chemical groups (e.g., thiols) as well as biochemical functions. Similarly, the six polymers chosen display substantial differences in their physicochemical properties, such as surface free energy, chemical composition, charged groups, and also microcrystallinity and microphase separation.

Radiolabeled proteins were used in parallel measurements to determine adsorbed protein surface concentrations.

Materials and Methods

Protein Isolation, Treatment and Character, za ti on

Canine proteins were isolated from citrated canine plasma. Human (HSA) and canine serum albumin (CSA) were obtained by a modifi­cation of the method of Day et al. [5], by affinity chromatography on fine Affi-gel blue (Bio-Rad). The protein, after concentration and buffer change by dialysis, was chroma­tographed on a fine Bio-gel P-100 column to separate the monomer. The monomeric protein

766

was then snap-frozen in a dry ice-ethanol bath and stored at -70°C prior to use. Human (HFGN) and canine fibrinogen (CFGN) were isolated by a modification of the method of Jakobsen and Kieruff [11], bye-alanine precipitation. The FGN obtained had a clottability of 98% or higher by the method of Coller [4]. Human (HFN) and canine fibronectin (CFN) were obtained by a modification of the method of Ruoslahti et al. [25], by affinity chroma­tography on gelatin-agarose (Bio-rad). The proteins were radio-labeled with 1251 using a modified Chloramine-T method (Iodobeads, New England Nuclear). The radio-iodinated protein was separated from free iodine on a desalting column of Bio-gel P-30 (Bio-rad). Radio­iodination levels were less than or equal to 0.13 mole of 1251 per mole of labeled protein.

All the proteins were restored to phosphate buffered saline (PBS) pH 7.39, containing 0.02% NaN3 and no divalent cations, in the final step before snap-freezing and storage. The purity of all proteins was established by polyacrylamide gel electro­phoresis in the presence of sodium dodecyl sulfate (SOS-PAGE), by the method of Weber and Osborn [27].

Adsorption of the Proteins to Test Polymers Polymers. The polymers used were low

density polyethylene (PE) (Intramedic), chromic acid-oxidized PE (OXPE), Teflon FEP® (FEP) (Cole Parmer), solution grade Biomer® polyetherurethaneurea (Biomer) (Ethicon), and two laboratory synthesized zwitterionic poly urethanes. Zwitteri onomer ZW is a polyurethaneurea made from polytetramethylene oxide (PTMO, MW 1000), 4,4'-diphenylmethane­diisocyanate (MDI), and N-methyl diethanolamine (MDEA), in a 1:3:2 ratio. Zwitterionization is accomplished by sulfonation of the tertiary amine on MDEA with 1,3-propanesultone [12]. Zwitterionomer ZS is a polydimethyl siloxane polyurethaneurea chain extended with MDEA and sulfonated with propane sultone. Details of the syntheses and characterization of these zwitterionomers are published elsewhere [12]. OXPE was freshly made by the chromic acid oxidation procedure of Rasmussen et al. [22], and possesses carboxylic groups on the surface in addition to ketone and aldehyde functions. The polyurethanes (Biomer, ZW, and ZS) were coated onto OXPE using the procedure of Lelah et al. [12].

Protein Adsorption. The method of preparation of the tubing surfaces and of protein adsorption is described here. Five cm segments of 0.318 mm ID tubing of the polymers were connected with 0.264 mm ID Silastic con­nectors (Dow Corning), and three-way stopcocks were attached at both ends of the resulting shunt. The shunts were washed with 0.125% Ivory detergent, copiously rinsed with double distilled deionized water, rinsed and filled with the adsorption buffer (PBS with 0.02% sodium azide), and equilibrated for 18 hrs at 22°C, under water.

For protein adsorption, the shunts were

Immunogold Labeling - Visualization of Adsorbed Proteins

held vertically in plexiglass racks and radiolabeled, prefiltered protein solutions were introduced from below with a syringe at a rate of 32 ml/min. After various pre­determined adsorption times, the bulk protein solution was flushed out at 100 ml/min with 20 shunt volumes of PBS. Care was taken to avoid the introduction of air-bubbles at all stages.

Immunogold Labeling. Immunogold labeling of the adsorbed proteins was performed by a modification of the method of Park et al. [17]. The adsorbed protein molecules were fixed~ situ using a 7 min. incubation with 1% glutaraldehyde in PBS, followed by PBS rinsing. The excess glutaraldehyde was neutralized by 20 min. incubation with 50 rrM glycine in PBS. The tubing was rinsed again with PBS. Aoproximately 3 mm sections of the prefixed protein-coated tubing segments were cut using a sharp double-edged blade, and transferred onto a ridged 96-well cover plate (Costar), without drainage. A 50 ul suspension of the immunogold particles (A525 = 6.2) was added and incubation was allowed for 45 min., at room temperature (ca. 22-24°C). With the known 30 ul volume of the section, the staining conditions were equivalent to incubation of the polymer surface with 1~mmunogold suspension of A5?5 = 3.88 (or 3xl0 particles/ml [13]). The section was then rinsed carefully and slowly in PBS, and transferred into a vial containing 2% glutar­aldehyde. After at least 24 h incubation, the section was prepared for electron microscopy. A protein-specific antibody and a non-specific antibody were always used. The non-specific staining of bare polymers was also examined. The cross-reactivities of the antibodies with non-homologous proteins was also tested. These controls were performed with each experiment.

A 2 cm segment of the prefixed protein­adsorbed tubing was cut, drained, and used for gamma counting in a Beckman 5500 gamma counter. The exact length of the segment was measured with a micrometer, and was used to determine the surface concentration of the adsorbed protein. Al cm piece from each end of each polymer segment was always cut out and discarded.

Table l

Immunogold Suspensions. The immunogold suspensions were prepared as previously reported [13]. Affinity-purified goat anti­human protein antibodies were used in all cases (Cooper Biomedical). The colloidal gold particles of an average diameter of 18 nm were prepared by reduction of boiling 0.01% HAuC14 with 1% trisodium citrate. The colloid was cooled, filtered (0.45 um), and pH adjusted (pH 7.3) just before use. The minimum amount of antibody necessary to stabilize the colloid (ca. 8-10 ug/ml colloid) was determined from adsorption isotherms as described by Horisberger and Rasset [9]. This calibration was found to be linear over the range of con-centrations from A525 = -0 to 20.

SEM Preparation; Samples were dehydrated in a graded ethanol series and were dried by the critical point method using molecular sieve-dried CO2 as the transitional fluid. The samples were sputter-coated with 10 nm gold and examined in a JEOL JSM 35C scanning electron microscope using 10-20 kV acceleration voltages and at 20,000 magnification.

Results

Control colloidal antibody preparations (i.e., antibodies non-specific for the protein tested) did not adsorb to any of the protein­coated surfaces indicating that the observed staining was due to specific antibody-protein binding and not due to non-specific antibody­protein interactions. Also antibody-gold was shown not to bind to uncoated surfaces. The effect of the surface on the adsorbing protein would be expected to be maximal in the protein layer closest to the surface. Thus adsorption conditions were so chosen as to obtain adsorbed surface concentrations in the monolayer range. Table 1 lists the monolayer surface con­centrations calculated from the dimensions of the protein molecules assuming random packing (with a packing coefficient of 0.555 [8]) and assuming that the adsorbed protein molecules were randomly oriented. Literature data on diffusivity and intrinsic viscosity were used to calculate the dimensions of human albumin

Monolayer Surface Concentrations of Adsorbed Proteins

Protein

Serum Albumin Fibrinogen Fibronectin

Protein Molecule Dimensions*

nm3

4.2x4.2xl4.l [22] 9.0x9.0x46.0 [23]

10.3xl0.3x66.0 [24]

Monolayer Surface Concentration

u g/cm2

0.29 0.30 0.29

Protein Molecules Per Immunogold particle 0

70 13 9.6

* The dimensions, represented as those of equivalent prolate ellipsoids are calculated from the diffusivity and intrinsic viscosity data given in the references in parentheses, for human proteins.

0 These approximate values are calculated based on the maximum immunogold coverage by 18 nm immunogold particles of 400 per um2, and the monolayer concentration of the adsorbed protein.

76 7

K. D. Murthy et al.

[20), fibronectin [28) and fibrinogen [7]. The adsorbed protein concentrations for

(CFGN) were in the monolayer range on FEP and ZW, but much higher on PE, OXPE, SB and ZS [16) at 60 minutes and beyond of adsorption from a 0.3 mg/ml solution concentration. For (CFN) all surfaces were within monolayer range [16) at 30 minutes of adsorption and beyond from a 0.07 mg/ml solution concentration. In the case of (CSA), the adsorbed protein concentrations were in the monolayer range on FEP, PE and ZS, but were much higher on OXPE, SB and ZW at 120 minutes of adsorption and beyond from a 0.90 mg/ml solution concentration [16).

Figure 1 CFGN adsorbed to OXPE for 60 minutes from a solution concentration of 0.~ mg/ml (surface concentration= 0.68 µg/cm) followed by labeling with anti-FGN colloidal gold part i c 1 es. Bar = 1. 0 µ m ( 1 7 kV)

Figure 3 HSA adsorbed to PE for 120 minutes from a solution concentration of 0.9 mg/ml (surface concentration= 0.27 µg/cm2) followed by labeling with anti-SA colloidal gold particles. Bar= 1.0 µm (20 kV)

768

Immunogold-labeled HFGN and CFGN appeared to cover all surfaces in a homogeneous fashion (Figure 1). Immunogold labeled human and canine albumin covered surfaces unevenly. On (Figure 3) PE, OXPE, SB and FEP the surface was completely covered. On ZS (Figure 2) and ZW only about half of the surface was covered. Large bare patches, on the order of 10 µm across and containing very few immunogold markers, occupied the rest of the surface. On the protein-coated areas of these polymers albumin labeling density was slightly lower. CFN and HFN deposited in the same dense, evenly distributed fashion as FGN (Figure 4). On

Figure 2 HSA adsorbed to ZS for 120 minutes from a solution concentration of O.~ mg/ml (surface concentration= 0.21 µg/cm) followed by labeling with anti-SA colloidal gold particles. Bar = 1.0 µm (15 kV)

Figure 4 HFN adsorbed to PE for 30 minutes from a solution concentration of 0.07 mg/ml followed by labeling with anti-FN colloidal gold particles. High magnification showing an area of homogeneous distribution. Bar= 1.0 µm (20 kV)

Immungold Labeling - Visualization of Adsorbed Proteins

Figure 5 HFN adsorbed to PE for 30 minutes from a solution concentration of 0.07 mg/ml followed by labeling with anti-FN colloidal gold particles. Low magnification demonstrating areas of networks, upper left, and homogeneous distribution, lower right. Bar= 1.0 µm (20 kV)

other areas of these polymers, CFN and HFN appeared to deposit in a regular network with bare spaces of up to 1 µmin diameter (Figures 5 and 6). This pattern was observed on both critical point and rapidly frozen specimens.

Discussion

Immunogold labeling appears useful for the observation of surface distribution and reten­tion of antigenic activity of adsorbed protein molecules. However, there are certain constraints which can be evaluated from direct calculations (Table 2). The maximum number of

Figure 6 HFN adsorbed to PE for 30 minutes from a solution concentration of 0.07 mg/ml followed by labeling with anti-FN colloidal gold particles. High magnification showing an area of networks. Bar = 1.0 µm (20 kV)

antibody molecules which may be adsorbed onto an immunogold particle can be determined from the size of an antibody molecule [21], by assuming either side-on (i.e., flat on the surface of the gold particle) or end-on (i.e., standing on the surface) adsorption of the antibodies to the gold particle. The true number of antibody molecules per particle should lie between these limits. From the antibody dimensions, and the gold particle size, the apparent size of the antibody-coated gold particle can be determined. The maximum density of immunogold particles, gmax, that can adhere to a fully antigen-coated surface can be

Table 2

Characteristics of Immunogold Particles

Gold Bead An ti bodies Immunogold Maximum Immunogold Immunogold Size Per GB0 Size Surface Concentration Apparent

Theoretical* Observed+ Size nm nm nm #/µm2 #/µm2 #/µm2 nm

2 1 2 1 2

5 1 2 14.4 38.8 5500 2750 12 2 14 21. 4 45. 8 1652 337 15 3 22 24.4 48.8 1280 299 18 5 32 27.4 51. 8 1020 263 400± 10 42 30 14 90 39.4 63.8 500 174 50 39 186 59.4 83.8 222 100 210± 10 59.4

0 Antibody dimensions were assumed to be 9.9x4. 7x4.7 nm3 for the stem and 9.8x4.2x4.2 nm3 for the arms [21].

1 Calculated Assuming Side-On Antibody Adsorption. 2 Calculated Assuming End-On Antibody Adsorption. * Calculated Assuming Random Packing. + See Text.

769

K. D. Murthy et al.

calculated from the immunogold size, and assuming random packing of the immunogold particles. For a 50 nm gold particle size, a value of 222 (for side-on antibody adsorption) or 100 (for end-on antibodies on gold particles) immunogold particles per um2 can be calculated. The observed maximum value reported by Park et al. [17] is 210±10 particles/µm2, indicating that the antibodies are adsorbed onto 50 nm gold beads in a predominantly end-on fashion. The highest value observed by Park et al. was about 310±40 particles/um2 for 18 nm immunogold particles [17]. Saturation had not occurred at this level of staining. At maximum staining of fibrinogen-coated surfaces with anti-fibrinogen coated 18 nm gold particles, we have found a particle density of 400±10 per µm2. This value is consistent with the data of Park et al. [17], and is much closer to that predicted for end-on adsorption [263] than for side-on adsorption [1020] of the antibodies. Thus in going from 50 nm to 18 nm gold bead size, the antibody adsorption conformation is suggested to shift towards end-on. This may be the effect of the lower surface area of the smaller gold bead, which may not allow spreading of the antibodies. It suggests that using smaller sizes of the gold particles will result in predominantly end-on anti body attachment. It has been observed that smaller particles are more efficient for immunogold labeling [6]. If the antibodies are adsorbed side-on, and thus spread on the gold surface, their antigen­binding activity may be diminished. On the other hand, the end-on adsorbed antibodies would retain more activity if the Fab of the end-on IgG's face outward. The practical constraint in the use of small colloidal gold particles is placed by the resolution of the scanning electron microscope, which at present limits us to the use of 10 nm and larger gold particles.

The relatively large size of the immuno­gold particle creates steric constraints in the labeling of adsorbed protein molecules. Thus, the effective diameter of the antibody-coated gold particles with an 18 nm gold bead at the center is calculated to be 42 nm, based on the observed maximum immunogold density of 400±10 per um2. From this, and from the calculated values of monolayer surface concentrations of the different proteins, one can calculate how many molecules of a protein will be covered by a single immunogold particle (Table 1). We observe that one 18 nm immunogold particle can cover an area equivalent to about 9.6 fibro­nectin molecules, 13 fibrinogen molecules, or 70 serum albumin molecules. Hence, if staining were continued to its end-point, the whole surface would become completely covered with immunogold particles, provided that at least 10% of the adsorbed fibronectin molecules, 8% of the fibrinogen, or 1.4% of the adsorbed albumin molecules were immunochemically active. No information regarding the nature of surface distribution or retention of immunochemical activity may then be obtained. Thus, it is

770

necessary to stain only to an intermediate extent, allowing a fraction of the molecules to be labeled. In this case, the adsorbed protein molecules possessing a relatively high immunochemical activity will be preferentially labeled. The present experiments were conducted with these constraints in mind. No direct correlation of the staining density with the hydrophilic or hydrophobic nature of the surface alone is possible, since the observed trends depend also upon the nature of the protein. This is partly because immunogold particles can only stain the uppermost layer of the adsorbed protein, and if the protein concentration increases above that of a mono­layer, the immunogold concentration may not increase any further.

The formation of networks observed in the case of fibronectins (CPFN and HFN) may be indicative of inter-molecular association between the adsorbed protein molecules. We have carefully purified our proteins to electrophoretic homogeneity (> 97% purity). Thus, the network of beads seen on the surfaces is not an artifact due to multimers being present in the protein solutions used. It is tempting to speculate that the surfaces may catalyze ooltimerization of the protein molecules. Since both albumin [20] and fibro­nectin [15] contain highly reactive free thiols as well as exchangeable dithiol bridges in the molecule, it may be that multirnerization occurs by the formation of inter-molecular dithiol bridges. This could lead to the formation of networks of the adsorbed protein molecules. Staining would result in the formation of net­works of gol ct-beads. The possibility that these patterns may be caused or influenced by sample preparation must also be considered. However, the pattern development observed was specific to only certain proteins and polymers, and was seen in both critical point and freeze­dried preparations. While we are continuing to evaluate the effects of sample preparation, the pattern formation is an inherent character of certain protein and surface combinations.

Summary

Immunogold beads effectively allowed the visualization of surface distribution of adsorbed proteins. The labeling technique demonstrated that often the adsorbed species exhibit dis ti net patterns on surfaces. The type of pattern formation depends upon the protein and surface involved. The immunogold distribution, the type and extent of pattern formation combined with radiolabel counting are clues towards the retention or non-retention of antigenic properties of the protein upon adsorption.

Acknowledgements

The authors wish to acknowledge the partial support of this research through National Institute of Health grants HL-21001 and HL-24046.

Immunogold Labeling - Visualization of Adsorbed Proteins

References

[l] Andrade JD (1985) Principles of protein adsorption. In: Surface and Interfacial Aspects of Biomedical Polymers, V 2, Protein Adsorption, Andrade JD (ed), pp 1-80, Plenum Press, N.Y. [2] Baier RE (1972) The role of surface energy in thrombogenesis. Bull. N.Y. Acad. Med. 48, 257-260. [3] Baier RE, Dutton RC (1969) Initial events in interactions of blood with a foreign surface. J. Biomed. Mater. Res. 3, 191-206. [4] Coller BS (1979) Asialofibrinogen supports platelet aggregation and adhesion to glass. Blood 53, 325-332. [SJ Day JF, Thornburg RW, Thorpe SR, Baynes JW (1979) Nonenzymatic glucosylation of rat albumin. J. Biol. Chem. 254, 9394-10004. [6] DeMey JR (1983) Thepreparation of immunoglobulin gold conjugates (!GS reagents) and then use as gold markers for light and electron microscopy in immunocytochemistry. In: Immunohistochemistry, Cuello AC (ed), pp 347-372. Wiley-Interscience. [7] Doolittle RF (1973) Structural aspects of the fibrinogen to fibrin conversion. Adv. Prot. Chem. 27, 1-109. [8] Feder J~Giaever I (1980) Adsorption of ferritin. J. Coll. Interf. Sci. 78, 144-148. [9] Horisberger M, Rosset J (1977T Colloidal gold, a useful marker for transmission and SEM. J. Hi stochem. Cytochem. 25, 295-305. [10] Ikada Y, Suzuki M, Tamada Y (1984) Polymer as Biomaterials, Shalaby S, Ratner BD, Hoffman AS, Horbett TA (eds), pp 135-147, Plenum Press, NY. [11] Jakobsen E, Kieruff P (1973) A modified B-alanine precipitation procedure to prepare fibrinogen free of antithrombin III and plasminogen. Thromb. Res. 3, 145-159. [12] Lelah MD, Grasel TG, PTerce JA, Cooper SL (1986) Ex vivo interactions and surface propertyrelationships of polyetherurethanes. J. Biomed. Mater. Res. 20, 433-468. [13] Loftus JC, AlbrechtRM (1983) Use of colloidal gold to examine fibrinogen binding to human platelets. Scanning Electron Microsc. 1983; IV: 1995-1999. [14] Macritchie F (1978) Proteins at interfaces. Adv. Prot. Chem. 32 283-326. [15] Mosher DF (1981) Fibronectin. Prog. Hemost. Thromb. 5, 111-151. [16] Murthy KD (T986) Protein adsorption and immunogold staining on polymeric biomaterials. M.S. Thesis, Univ. of Wisconsin-Madison. [17] Park K, Albrecht RM, Simmons SR, Cooper SL (1986) A new approach to study adsorbed proteins on biomaterials. J. Coll. Interf. Sci. 111, 197-212. [18] Park K, Mosher DF, Cooper SL (1986) Acute surface-induced thrombosis in the canine ex vivo model: Importance of protein compos,tion of the initial monolayer and plate­let activation. J. Biomed. Mater. Res. 20, 589-612. -[19] Paytner RW, Ratner BD, Horbett TA, Thomas HR (1984) XPS studies on the organization of

771

adsorbed protein films on fluoropolymers. J. Coll. Interf. Sci. 101, 233-245. [20] Peters Jr TH (198-;-r- Serum albumin. Adv. Prot. Chem. 37, 161-245. [21] Pilz I,Puchwein G, Kratky 0, Herbst M, Haager 0, Gall WE, Edelman GM (1970) Small angle x-ray scattering of a homogeneous yGI immunoglobulin. Biochem. 9, 211-214. [22] Rasmussen JR, Stredonsky ER, Whitesides GM (1977) Introduction, modification and characterization of functional groups on the surface of low density polyethylene. J. Am. Chem. Soc. 99, 4736-4740. [23] RatnerBD, Horbett TA, Shuttleworth D, Thomas HR (1981) Analysis of the organization of protein films on solid surfaces by ESCA. J. Col 1. Interf. Sci. 83, 630-643. [24] Rudee ML, Price TW-(1985) The initial stages of adsorption of plasma derived proteins on artificial surfaces in a controlled flow environment. J. Biomed. Mater. Sci. _!i, 57-66. [25] Ruoslahti E, Hayman EG, Pierschbacher M, Engvall E (1982) Fibronectin: purification, immunochemical properties and biological properties. Meth. Enzymol. 82, 803-831. [26] Seaman GVF (1983) Plasmaprotein interac­tions at biological interfaces. Thromb. Res. (Supp. V), pp 83-91. [27] Weber K, Osborn M (1969) The reliability of MW determinations by dodecyl sulfate­polyacrylamide gel electrophoresis. J. Biol. Chem. 244, 4406-4410. [28] Williams EC, Janmey PA, Ferry JD, Mosher DF (1982) Conformational states of fibro­nectin. J. Biol. Chem.~. 14973-14978.

Discussion with Reviewers

L. Vroman: Phosphate buffer is a detergent (and was so described by Giaever at the ACS meeting in Anaheim-1986). Could this account for the incomplete coverage of proteins and for the finding that antibody-gold did not bind to uncoated surfaces? Authors: The antibodies are essentially irreversibly adsorbed on gold sols and maintain their ability to interact with antigens. Therefore, immunogold particles can bind to protein molecules adsorbed on a surface, if the gold sol carries antibodies specific to the particular protein. Since fibrinogen, fibro­nectin and serum albumin samples were treated the same, patterns formed were very distinct. Antibody gold does not bind to uncoated surfaces because the antibody-antigen complex only recognizes adsorbed protein molecules.

L. Vroman: Is Ivory detergent a soap? Does ,t not leave a film of fatty acids? Authors: The "Ivory" we used is a detergent and does not contain fatty acid groups.

L. Vroman: Effects of time on surface concentration of proteins must have been obtained from radiolabel counts, since gold labeling was intentionally performed only briefly and incompletely to reveal structure.

K. D. Murthy et al.

Data should be tabulated, mainly since the authors do not show whether optima or maxima were obtained. We have shown that certain proteins, not only in plasma but as solutions of purified proteins, are adsorbed with an optimum depending on time, concentration and (if present) gap width between adsorbing surfaces! Reviewer II: What were the adsorption values for fibronectin? Authors: The following table indicates the surface concentration values of the protein/ surface combinations studied. Values are for canine plasma proteins ( text reference [16]).

Protein Surface Concentrations {µg/cm2)

Surface Fibrinogen Fibronectin Serum Albumin

PE FEP SB

D XPE zw ZS

D. 42±0. 01 0. 29±0. 02 0.87±0.03 0. 68± 0. 06 0.26±0.09 0. 60±0. 06

0.21 ±0.03 0.23 ±0.03 0.16 ±0. 01 0.19 ±0. 00 0. 018±0. 013 0.15 ±0.02

0.13±0.01 0. 12± 0.01 0.47±0.02 1.01±0.15 0. 56±0. 01 0. 15± 0.01

Note: Fibrinogen adsorbed for 60 minutes at D. 30 mg/ml. Fibronectin adsorbed for 30 minutes at 0.07 mg/ml. Serum albumin adsorbed for 120 min. at 0.90 mg/ml. The values are for (n=3)±standard deviation.

L. Vroman: Retention of antigenicity of adsorbed protein can only be assessed by comparing with an independent other method. presume the authors are referring to their radiolabel counts, but they are not given. Authors: We do refer to radiolabeling counts, the above Table shows the surface concentration values for each protein/material combination.

L. Vroman: Referring to Figures 4, 5 and 6. Are all of these taken from one sample? Authors: Yes, this is the same sample. We found several types of patterns to exist on the same surface.

Reviewer II: The chloramine-T ioctination method ,swell known to alter (drastically at times) protein properties. This problem has been described several times in the literature, for fibrinogen especially. In the case of fibronectin, it destroys the proteins ability to bind to gelatin. Did you ever test to see if your 125I proteins preferentially adsorbed or if their function (fibrinogen clottability, fibronectin gelatin binding) changed? Authors: Yes, we tested the preferential adsorption of fibrinogen with different ratios of labeled to unlabeled protein. No increase in adsorption was observed. The preferential adsorption of the same labeled species has been shown to be substrate (system) dependent [29,30]. Also work by Young [33] has shown that there is no preferential adsorption in studies conducted in our laboratory using the same experimental procedure as described here. Fibrinogen clottability did not change significantly after radio-iodination [31].

772

Reviewer II: Was your Affi-gel blue purified albumin really pure? This matrix binds lots of IgG, too, which co-elutes with the albumin. I believe this is well known now, even to the manufacturers, who originally touted this stuff as pretty specific for albumin. I always have to do a second, gradient elution on DEAE cellulose to get clean albumin from the Affi­gel blue eluate. Authors: After elution from a Affi-gel blue column the following steps were taken to further purify albumin. 1. Gradient elution on DEAE Sephadex (Sigma). 2. Separation of monomer on a Bio-gel P-150 column. 3. Gel electrophoresis in the presence of sodium dodecyl sulfate to assure that the monomer was separated from multimers.

Reviewer II: It sounds like you are saying human and canine fibrinogen have substantial differences in their molecular weights etc. but I think you mean that albumin and fibrinogen etc. differ a lot. Which do you mean? Authors: The molecular weights of fibrinogen (340 kd) and serum albumin (68 kd) differ.

Reviewer II: Aren't you worried that glutar­aldehyde prefixing will change the antigen binding site so the antibody won't bind or bind as well? Authors: Glutaraldehyde prefixing was used to fix the protein on the surface. We have conducted similar experiments without prefixing and have found there is no difference in antibody binding.

Reviewer II: Are your "affinity-purified" antibodies pure IgG? Lots of manufacturers add albumin back to stabilize them, I think. Did you run them on gels? Authors: The antibodies were separated from other serum proteins using a DEAE-Sephacel (Sigma) column equilibrated with 0.05 M Tris­buffer pH 8.5. Gels were run on the IgG's and results indicated that they were void of serum proteins.

Reviewer II: I don't understand what the sentences beginning "The minimal amount••••20." are supposed to be telling me. Does all this mean you saturated the gold surfaces with antibody? Why do you need a calibration curve? Authors: Yes, during preparation of the 1mmunogold complex, if the protein was not adsorbed and the gold particles were not stabilized, aggregation of the gold granules was indicated by a color change from red to light blue (text reference [17]). The minimum amount of protein to prevent this color change was used. The calibration curve was used to determine the concentration of immuno­gold solutions by measuring absorbance at 525 nm [32] (text reference [9]).

Reviewer II: I think at least one SEM documenting visually the lack of non-specific immunogold uptake by bare and protein coated surfaces that you describe would help a lot. Authors: SEM micrographs documenting the lack of non-specific immunogold uptake have been

Immunogold Labeling - Visualization of Adsorbed Proteins

published previously and are shown in text reference [ 6].

Reviewer II: When you've first glutaraldehyde fixed, then immunogold treated, then fixed, ethanol and CO2 dried, and sputter coated with still more gold, do you really have confidence that none of the myriad places artifacts could enter did not do so? Why? Any good controls on all this? Eberhart concluded networks were drying/retraction artifacts reflecting differences in the protein/surface and protein­protein bonding with different proteins and different surfaces. Why don't you consider this? Authors: We rapidly froze samples (sample temperature -70°C) without fixation and found that the same type of networks form on the polymer/protein combinations. Freeze drying at -70°C is well below the collapse temperature of the protein film. Controls were run with each experiment; critical point drying and freeze drying techniques showed the same results. Si nee the two types of drying used are different in principle and since identical patterns are formed using each method, we do not consider drying/retraction.

Reviewer II: I find the entire discussion on sizes and numbers of particles one should expect very unlikely. How do you know that antibodies don't rearrange differently (e.g. "side-on" versus "end-on") depending on exactly what antibody concentration you use to adsorb? Unless adsorbed density and adsorbing conditions are constant between 50 nm and 18 nm in particles, its quite likely different results would be obtained. Do you know the effect of antibody loading of the immunogold particles on their binding to antigen coated surfaces? This seems likely to be very important and I'm not at all confident your whole technique and calculations aren't merely artifacts of exactly how much antibody you loaded on the surfaces. Authors: Table 2 shows the theoretical calculations of the number of antibodies adsorbed to various size goldmarkers. What we are saying is that the number of antibodies arrange differently (e.g. side-on or end-on) depending on the pH/concentration isotherms and the actual conditions such as temperature and species. Below a minimum concentration, the bead is not completely covered and aggregation occurs in buffer containing free ionic species. Above this point full coverage occurs and there is steric stabilization of the gold complex against Yan der Waals aggregation. Once enough protein is present for stabilization, increasing the antibody concentration will increase the antibody covering beyond a monolayer. Concentration influences the thickness of the antibody layer as determined by light scattering.

773

Additional References

[29] Grant WH, Smith LE, Stromberg RT (1977) Radiotracer techniques for protein adsorption measurements. J. Biomed. Mater. Res. Symp . .§_, 33-4 7. [30] Hennink WE, Dost L, Feijen J, Kim SW (1983) Interaction of albumin-heparin conjugate preadsorbed surfaces with blood. Trans. Am. Artif. Intern. Organs 29, 200-205. [31] Lelah MD, Lambrecht LK, Cooper, SL ( 1984) A canine ex vivo series shunt for evaluating thrombus depos, tion on polymer surfaces. J. Biomed. Mater. Res. 18, 475-496. [32] Park K, Simmons~R, Albrecht RM (1987) Surface characterization of biomaterials by immunogold staining. Scanning Microsc. 1(1), 339-350. [33] Young BR, Pitt WG, Cooper SL (1987) Protein adsorption on polymeric biomaterials: I. Adsorption isotherms. Submitted to J. Colloid Interface Sci.